Contemporary school-based smoking prevention programs that combine refusal skills training with techniques to enhance general personal and social competence can be highly effective in reducing adolescent smoking. While etiology research has shown that social resistance skills are important in keeping young people away from smoking, little work has focused on how social competence (e.g., assertiveness, communication skills) and personal competence (e.g., cognitive and behavioral self-management skills) are protective in terms of youth smoking. The etiological mechanisms remain particularly unclear among youth in high-risk settings. Furthermore, national survey data show that prevalence rates for adolescent smoking differ according to ethnicity, gender, and geographic location. Thus it is important to determine if competence-based etiological models can account for the initiation and escalation of adolescent smoking among different subgroups of adolescents. A primary goal of the proposed research is to develop, test, and refine several etiologic models that focus on the role of social and personal competence skills in adolescent cigarette smoking, and to examine these models among two longitudinal samples of middle school students: a largely white, suburban sample and a predominantly minority, inner-city sample. This goal will be accomplished through secondary analysis of untreated control students who participated in one of two school-based drug abuse prevention trials. The proposed research aims to elucidate how competence skills influence the initiation and escalation of cigarette smoking during the critical middle-school years. For example, mediational analyses will examine the roles of affective self-regulation and perceived social benefits of smoking as factors that explain how poor competence leads to youth smoking; and moderational analyses will examine whether good competence skills buffer the effects of other salient risk factors for tobacco use. The hypothesized models will be cross-validated among subgroups of the two longitudinal samples including by ethnicity and gender. Because different etiologic factors may contribute to varying levels of cigarette smoking, prediction of experimental versus heavy smoking will be investigated among subgroups of students. Several multivariate statistical techniques will be used including multiple linear and logistic regression, structural equation modeling, and latent growth modeling. The long-term goals are to improve our knowledge regarding the development of tobacco use among youth of different backgrounds, and to ultimately refine and improve smoking prevention programs for diverse youth populations.